Joys of standard
Posted: 2018/10/12 15:49:37
You know the mantra:
No compiler is standard compliant but some compilers are more standard compliant than others.
And you evolve faith on proprietary compilers being "less" than the faithful GCC.
Then you write code, without checking the actual standard, and everything just "works"
on CentOS, as it "should". But the evil reality demands a port to a proprietary OS.
Worse than that, a use of proprietary (but no-cost, stripped version) compiler suite.
Your faith is strengthened by the observation of plenty of trouble and woes. Until you find the culprit.
Two declarations:
One is described in all documentation about the standard that one can easily stumble on,
and the other -- unfortunately -- can be found from GCC.
"Just an another bug-hunt."
No compiler is standard compliant but some compilers are more standard compliant than others.
And you evolve faith on proprietary compilers being "less" than the faithful GCC.
Then you write code, without checking the actual standard, and everything just "works"
on CentOS, as it "should". But the evil reality demands a port to a proprietary OS.
Worse than that, a use of proprietary (but no-cost, stripped version) compiler suite.
Your faith is strengthened by the observation of plenty of trouble and woes. Until you find the culprit.
Two declarations:
Code: Select all
template <class T>
T
valarray<T>::operator[] ( site_t ) const;
template <class T>
const T &
valarray<T>::operator[] ( site_t ) const;and the other -- unfortunately -- can be found from GCC.
"Just an another bug-hunt."